I am experiencing random unexplained system crashes while playing Last Epoch. Without any prior noticeable warning, the computer instantly turns off, and then starts up again a few moments later. Windows error logs show an unexpected shutdown but no errors preceding it. This may however not be a problem with Last Epoch itself. Further explanations below. DxDiag.txt (111.4 KB) le_graphicsmanager.ini (474 Bytes)
When I first received this PC, I almost immediately started having this problem. After some research, it seems to be a relatively common issue with the RTX 30-series, especially the 3070. However, the problem eventually went away for no known reason, and have been gone for about six months.
After 0.8.3, I had another one of these crashes occur after about 2.5 hours. Since then, Iām getting a crash every 2-5 hours of playtime with exactly the same symptoms. I have yet to see if the crashes also happen without Last Epoch running, since Iām playing it obsessively, but if it happens without LE running Iāll update the thread.
Iām posting this because Aluxaeterna asked me to on discord, despite the fact that LE is likely not the root cause behind the crashes. 0.8.3 definitely seems to have triggered the symptoms again though.
Looking at the info, your system spec should obviously be more than adequate to play LE.
Your drivers and OS look up to date (windows has one more patch to 19043) so thats unlikely to be the cause of anything as serious as you describe - provided the installations are ok and there are no physical hardware issues.
Looking at the dxdiag diagnostics (end of the file) there are a LOT of errors that have nothing to do with LEā¦ some are pretty serious and should be traced and resolved else you are likely to have these issues indefinitely.
LiveKernelEvent 141 is a hard one as it could literally refer to any number of errors from hardware, thermals, system files, driver issues etcā¦ Usually when this error happens, your system will just reboot with no apparent reason and it tends to happen when you are stressing the system (like playing games) but not always.
ROGLiveService.exe & ArmourySocketServer.exe there are quite a few of these errors and doing a quick Google shows that there are LOTS of issues with these applications & services that cause crashes on users systems. Some seem related to bad dlls / installs / old versions & others seem to be conflicts with other software like EasyAntiCheatā¦
My recommendation would be:
Update your Windows to 19043.
Run some stress testing on your system using things like Prime95/Furmark/Heaven or whatever - monitor your system temps, components, system power while doing thisā¦ and intend to run the tests for similar times as you do when you experience the crashes while playing gamesā¦ dont leave things unattended when you do. You may even want to do a hardware/bios/memtest on you 64gb ramā¦ if one stick is not running well, it could explain things.
Resolve the ROG & Armoury issues - either by uninstalling or trying newer versions. These could potentially be one of the main causes of crashesā¦
Because of the unexplained livekernel eventsā¦ I would run Windows System file checker to make sure that your OS system files are okā¦ Its unlikely to do anything but its worth trying when you have odd errors like this. Googling the livekernel event is likely to give you a headache but Iād browse the first few results or expand the search to try and see if any of the results trigger something that may apply to your setup - only you would really know.
I would recommend that you double check all the hardware drivers in your system are up to date and working correctly. This includes things like BIOS updates. You may even want to do a safe mode GPU driver installation (or use something like DDU) to ensure that your GPU drivers are correctly installed.
In line with the ROG issuesā¦ Double check any other applications that you use on your system - especially those that are automatically running or are āalways availableā - there is always the possibility that you are running some service or application that might seem irrelevant but is actually buggy and doing something that is causing problemsā¦ If there is any regularity in the timing of your crashes/issues then look at things that have timed action - like an app that checks for updates every few hours e.gā¦
First off, thank you for such a detailed reply! Iām very impressed. Hereās an update:
Update done.
Stress test using furmark showed no issues whatsoever for a 2 hour run. Temperature remained steady at 63 degrees using full AA. System barely broke a sweat.
Uninstalled ROG and Armoury. No similar errors in the logs now.
System file checker found something corrupt and repaired it.
Hardware drivers updated, including a fresh install of the GPU driver. Motherboard BIOS also updated.
Investigating the timing of the crashes did produce interesting results. The times are as follows: 21:28, 22:56, 18:54, 23:03, 00:28, 15:56. Thatās the log entries of windows was not properly shut down, meaning after the computer had rebooted I believe. The fact that theyāre all Ā± 10 minutes of half hour steps makes me hella suspicious. I have turned off all scheduled tasks for the moment to see if that makes a difference.
Another crash happened before I turned off the scheduled tasks (6), so the other steps didnāt help.
Turning off scheduled tasks didnāt help, nor did reseating the graphics card, and providing it with two separate power cables instead of one daisy-chained one. This crash at around 18:50.
I just ran Furmark stress test after a fresh boot, the crash happened after a while. Last Epoch was not running and had not been running that session.
I donāt know if itās a coincidence that the crashes returned after months of inactivity right after 0.8.3 dropped (I played last epoch on 0.8.2 in the week prior without issue), or if playing 0.8.3 triggered some software change that made the crashes start happening again, but either way I doubt itās something EHG can fix.
Sounds like you did some good systematic testing thereā¦
Furmark crashing just confirms that a stress point specifically on your GPU existsā¦ Something that LE can easily do with the settings you are trying to run it atā¦ I dont know what other stress tests you ran before, but for example Prime95 would only stress your CPU not your GPUā¦
Furmark is obviously not real world and is intended to try and overload a GPU to test stability/thermals/power in a worst caseā¦ Were you monitoring GPU usage/temps/power draw when it crashed? That could give you a clue as to what your GPU limits might beā¦ Maybe the GPU draws too much power after a while (e.g. PSU issue)ā¦ maybe it thermal throttles/crashes (Case thermals, fan curves, problem with the GPU fan/cooling)ā¦
The issue about the long period of inactivity being related could be anythingā¦ you probably had a few windows patches or driver updates or even installed new software and that could be involvedā¦
As a quick testā¦ (not sure if you have already tried it)ā¦ Load up LE and set the in-game settings to very low and every feature to disabled. change the resolution to 1080p Fullscreen display mode with the framerate limit set to 60fpsā¦ Monitor your GPU while playing and see if it breaks a sweatā¦ it shouldnt and the game might not reach the āFurmarkā level and be more stableā¦
Considering that I just had a crash shortly after boot, with nothing running except Brave browserā¦ Iām gonna go ahead and get in touch with the warranty people for my computer.
Definitely sounds like its getting worse and probably hardware related. If it is hardwar, it could be anything from motherboard to power supply to a flaky gpu and anything inbetween so unless you have spare parts to test going the warranty route is probably the simplestā¦
Before totally going the hardware warranty route, you may want to try a clean OS install and just make 100% its not some really wonky drivers or system files etc.
Iām very fortunate to have a very comprehensive warranty route to go down. Itās one of the perks of using the company I did, they put together the parts I selected and give me warranty on all the parts in the build for 3 years.
Monitoring the GPU stats, the only thing of note is that it occasionally spikes power usage/other stats to a very high value for a short period. This doesnāt cause a crash in the times Iāve seen it, but itās possible that the times it causes a crash donāt make it into the logs.
A clean install hasnāt helped either, unfortunately.
